Introduction

Here we are analyzed about GSE114260 dataset, detailed information about this dataset is given below.

Platform title: Illumina HiSeq 2000 (Homo sapiens)

Submission data: Nov 02 2010

Last update data: Mar 27 2019

Organism: Homo sapiens

Number of GEO datasets that use this techology : 7897

Number of GEO samples that use this technology : 122103

Refer to corresponding paper, combined in file name means this file contains samples from both human and mice, since we only care about human, don’t need to use the two combined file. Therefore, the stranded_read_counts is the data file of interest.

To clarify, ER in this data set stands for estrogen receptor, not Endoplasmic reticulum.In this dataset, we have 2 groups with 3 replicates each, so 6 samples in total. C4-12ERaERE (ER lacking cell line stably transfected with ERaERE) relative to the MCF7 cells were used for RNAseq analysis. The treatment group undergo paroxetine and estrogen (E2) treatment, while in control group there is only paroxetine treatment.

Raw data has been normalized using trimmed mean approaches. The we performed threshold over-representation analysis, by g:Profiler, on the normalized dataset. The enrichment analysis result support conclusion in the original paper.Top gene returned is ESR1 and top term returned is ‘GO:0006614 SRP-dependent cotranslational protein targeting to membrane’. Refer to gene summary provided by HGNC, ESR1 encodes an estrogen receptor. In addition, there are paper shows that term ‘GO:0006614 SRP-dependent cotranslational protein targeting to membrane’ is related to breast cancer. Since the orignal paper conclude that response of a certain drug for breast cancer is dependent on ER, the conclusion is supported.

top 10 gProfiler ORA result
source	term_name	term_id
GO:BP	SRP-dependent cotranslational protein targeting to membrane	GO:0006614
GO:BP	viral transcription	GO:0019083
GO:BP	cotranslational protein targeting to membrane	GO:0006613
GO:BP	viral gene expression	GO:0019080
GO:BP	viral process	GO:0016032
GO:BP	symbiotic process	GO:0044403
GO:BP	protein localization to endoplasmic reticulum	GO:0070972
GO:BP	nuclear-transcribed mRNA catabolic process, nonsense-mediated decay	GO:0000184
GO:BP	protein targeting to ER	GO:0045047
GO:BP	establishment of protein localization to endoplasmic reticulum	GO:0072599

Non-threshold Gene Set enrichment analyisis

method

Choose GSEA(version 4.0.3) preranked analysis(Subramanian et al. 2005) since here we are using a ranked gene list. Database used is from bader lab, publised April 1st, 2020(Merico et al. 2010). Gene set permutation is the default and only choice for preranked analysis in GSEA. As for minimum gene set size, default GSEA setting ,which is 15, remain unchanged. However, maximum size of gene set is reduced to 200 in order to reduce runtime. Number of permutation is 1000.

Result

Summary

Upregulated correspond to upregualted in treated samples and downregulated correspond to downregulated in treated sample.

In upregulated category, top gene returned is SELENOCYSTEINE SYNTHESIS%REACTOME%R-HSA-2408557.2, pvalue = 0.000, ES = 0.85, NES = 3.01, FDR = 0.000.

In downregulated category, top gene returned is HALLMARK_ESTROGEN_RESPONSE_EARLY%MSIGDB_C2%HALLMARK_ESTROGEN_RESPONSE_EARLY, pvalue = 0.000, ES = -0.77, NES = -2.69, FDR = 0.000.

Top 10 upregulated from GSEA

NAME	GS.br..follow.link.to.MSigDB	SIZE	ES	NES	RANK.AT.MAX	LEADING.EDGE
VIRAL MRNA TRANSLATION%REACTOME%R-HSA-192823.3	VIRAL MRNA TRANSLATION%REACTOME%R-HSA-192823.3	84	0.8643032	3.007869	1179	tags=74%, list=9%, signal=80%
PEPTIDE CHAIN ELONGATION%REACTOME%R-HSA-156902.2	PEPTIDE CHAIN ELONGATION%REACTOME%R-HSA-156902.2	84	0.8640424	2.994553	1179	tags=75%, list=9%, signal=82%
EUKARYOTIC TRANSLATION ELONGATION%REACTOME%R-HSA-156842.2	EUKARYOTIC TRANSLATION ELONGATION%REACTOME%R-HSA-156842.2	88	0.8542727	2.989950	1179	tags=72%, list=9%, signal=78%
SELENOCYSTEINE SYNTHESIS%REACTOME%R-HSA-2408557.2	SELENOCYSTEINE SYNTHESIS%REACTOME%R-HSA-2408557.2	87	0.8538864	2.988714	1179	tags=71%, list=9%, signal=78%
EUKARYOTIC TRANSLATION TERMINATION%REACTOME%R-HSA-72764.4	EUKARYOTIC TRANSLATION TERMINATION%REACTOME%R-HSA-72764.4	88	0.8575868	2.967159	1179	tags=70%, list=9%, signal=77%
CYTOPLASMIC RIBOSOMAL PROTEINS%WIKIPATHWAYS_20200310%WP477%HOMO SAPIENS	CYTOPLASMIC RIBOSOMAL PROTEINS%WIKIPATHWAYS_20200310%WP477%HOMO SAPIENS	84	0.8384490	2.946797	1179	tags=70%, list=9%, signal=76%
FORMATION OF A POOL OF FREE 40S SUBUNITS%REACTOME DATABASE ID RELEASE 72%72689	FORMATION OF A POOL OF FREE 40S SUBUNITS%REACTOME DATABASE ID RELEASE 72%72689	95	0.8257862	2.941885	1203	tags=69%, list=9%, signal=76%
COTRANSLATIONAL PROTEIN TARGETING TO MEMBRANE%GOBP%GO:0006613	COTRANSLATIONAL PROTEIN TARGETING TO MEMBRANE%GOBP%GO:0006613	92	0.8197061	2.934378	1179	tags=66%, list=9%, signal=72%
SELENOAMINO ACID METABOLISM%REACTOME DATABASE ID RELEASE 72%2408522	SELENOAMINO ACID METABOLISM%REACTOME DATABASE ID RELEASE 72%2408522	105	0.8209368	2.928979	1179	tags=61%, list=9%, signal=66%
SRP-DEPENDENT COTRANSLATIONAL PROTEIN TARGETING TO MEMBRANE%REACTOME%R-HSA-1799339.2	SRP-DEPENDENT COTRANSLATIONAL PROTEIN TARGETING TO MEMBRANE%REACTOME%R-HSA-1799339.2	106	0.8149124	2.921791	1179	tags=63%, list=9%, signal=69%

Top 10 downregulated from GSEA

NAME	GS.br..follow.link.to.MSigDB	SIZE	ES	NES	NOM.p.val	FDR.q.val	FWER.p.val	RANK.AT.MAX	LEADING.EDGE
HALLMARK_ESTROGEN_RESPONSE_EARLY%MSIGDB_C2%HALLMARK_ESTROGEN_RESPONSE_EARLY	HALLMARK_ESTROGEN_RESPONSE_EARLY%MSIGDB_C2%HALLMARK_ESTROGEN_RESPONSE_EARLY	173	-0.7737817	-2.686789	0.0000000	0.0000000	0.000	1704	tags=51%, list=13%, signal=58%
HALLMARK_ESTROGEN_RESPONSE_LATE%MSIGDB_C2%HALLMARK_ESTROGEN_RESPONSE_LATE	HALLMARK_ESTROGEN_RESPONSE_LATE%MSIGDB_C2%HALLMARK_ESTROGEN_RESPONSE_LATE	157	-0.6456295	-2.241304	0.0000000	0.0000000	0.000	1053	tags=31%, list=8%, signal=33%
PATHWAYS AFFECTED IN ADENOID CYSTIC CARCINOMA%WIKIPATHWAYS_20200310%WP3651%HOMO SAPIENS	PATHWAYS AFFECTED IN ADENOID CYSTIC CARCINOMA%WIKIPATHWAYS_20200310%WP3651%HOMO SAPIENS	56	-0.7533783	-2.235558	0.0000000	0.0000000	0.000	1273	tags=50%, list=9%, signal=55%
EYE MORPHOGENESIS%GOBP%GO:0048592	EYE MORPHOGENESIS%GOBP%GO:0048592	56	-0.6926590	-2.060055	0.0000000	0.0038307	0.014	1269	tags=25%, list=9%, signal=27%
VISUAL SYSTEM DEVELOPMENT%GOBP%GO:0150063	VISUAL SYSTEM DEVELOPMENT%GOBP%GO:0150063	111	-0.6208371	-2.058356	0.0000000	0.0030645	0.014	2187	tags=32%, list=16%, signal=38%
CAMERA-TYPE EYE DEVELOPMENT%GOBP%GO:0043010	CAMERA-TYPE EYE DEVELOPMENT%GOBP%GO:0043010	81	-0.6463426	-2.053742	0.0000000	0.0025538	0.014	403	tags=16%, list=3%, signal=16%
HISTONE MODIFICATIONS%WIKIPATHWAYS_20200310%WP2369%HOMO SAPIENS	HISTONE MODIFICATIONS%WIKIPATHWAYS_20200310%WP2369%HOMO SAPIENS	39	-0.7408906	-2.038358	0.0000000	0.0029178	0.019	1004	tags=36%, list=7%, signal=39%
EYE DEVELOPMENT%GOBP%GO:0001654	EYE DEVELOPMENT%GOBP%GO:0001654	110	-0.6154625	-2.036570	0.0000000	0.0029324	0.022	2187	tags=32%, list=16%, signal=38%
SENSORY ORGAN MORPHOGENESIS%GOBP%GO:0090596	SENSORY ORGAN MORPHOGENESIS%GOBP%GO:0090596	81	-0.6355954	-2.026007	0.0000000	0.0038485	0.032	1269	tags=22%, list=9%, signal=24%
SENSORY SYSTEM DEVELOPMENT%GOBP%GO:0048880	SENSORY SYSTEM DEVELOPMENT%GOBP%GO:0048880	115	-0.6104676	-2.022075	0.0000000	0.0036647	0.034	2187	tags=31%, list=16%, signal=37%
LXR-MEDIATED SIGNALING%REACTOME DATABASE ID RELEASE 72%9024446	LXR-MEDIATED SIGNALING%REACTOME DATABASE ID RELEASE 72%9024446	36	-0.7204237	-1.989345	0.0000000	0.0087138	0.089	824	tags=36%, list=6%, signal=38%
CAMERA-TYPE EYE MORPHOGENESIS%GOBP%GO:0048593	CAMERA-TYPE EYE MORPHOGENESIS%GOBP%GO:0048593	41	-0.7044234	-1.976283	0.0017182	0.0113972	0.128	403	tags=17%, list=3%, signal=18%
SENSORY ORGAN DEVELOPMENT%GOBP%GO:0007423	SENSORY ORGAN DEVELOPMENT%GOBP%GO:0007423	147	-0.5809889	-1.973475	0.0000000	0.0111504	0.136	2187	tags=29%, list=16%, signal=34%
HISTONE METHYLATION%GOBP%GO:0016571	HISTONE METHYLATION%GOBP%GO:0016571	56	-0.6673597	-1.964474	0.0000000	0.0124588	0.159	848	tags=29%, list=6%, signal=30%
PKMTS METHYLATE HISTONE LYSINES%REACTOME DATABASE ID RELEASE 72%3214841	PKMTS METHYLATE HISTONE LYSINES%REACTOME DATABASE ID RELEASE 72%3214841	42	-0.6924467	-1.962205	0.0000000	0.0118301	0.162	1004	tags=31%, list=7%, signal=33%
CELL-SUBSTRATE JUNCTION ASSEMBLY%GOBP%GO:0007044	CELL-SUBSTRATE JUNCTION ASSEMBLY%GOBP%GO:0007044	26	-0.7598497	-1.954675	0.0017271	0.0134554	0.191	1049	tags=38%, list=8%, signal=42%
REGULATION OF MECP2 EXPRESSION AND ACTIVITY%REACTOME%R-HSA-9022692.1	REGULATION OF MECP2 EXPRESSION AND ACTIVITY%REACTOME%R-HSA-9022692.1	28	-0.7356510	-1.953026	0.0000000	0.0133243	0.200	1185	tags=39%, list=9%, signal=43%
MUSCLE CELL DIFFERENTIATION%GOBP%GO:0042692	MUSCLE CELL DIFFERENTIATION%GOBP%GO:0042692	84	-0.6097430	-1.945164	0.0000000	0.0145124	0.228	1138	tags=25%, list=8%, signal=27%
CELLULAR GLUCOSE HOMEOSTASIS%GOBP%GO:0001678	CELLULAR GLUCOSE HOMEOSTASIS%GOBP%GO:0001678	37	-0.7022272	-1.938401	0.0000000	0.0157346	0.256	2153	tags=49%, list=16%, signal=58%
RETINA DEVELOPMENT IN CAMERA-TYPE EYE%GOBP%GO:0060041	RETINA DEVELOPMENT IN CAMERA-TYPE EYE%GOBP%GO:0060041	40	-0.6929891	-1.932622	0.0000000	0.0167860	0.282	403	tags=20%, list=3%, signal=21%

Comparision

Compare to result from thresholded over-representation analysis, there are several common terms.

Between GSEA upregulated result and ORA result:SRP-dependent cotranslational protein targeting to memebrane, cotranslational protein targeting to membrane, protein targeting to ER and establishment of protein localization to endoplasmic reticulum. Moreover, in ORA result, we have term ‘translational initiation’, while in upregualted GSEA result, we have ‘CAP-dependent translational initiation’ and ‘eukaryotic translation initiation’

No common terms in top 20 terms from downregulated GSEA and ORA result.

Also, we checked common gene related to both results. Number of common gene is given in venn diagram. Detailed list of common gene are listed.

This is a straight forward comparision.

Common Genes between GSEA result and ORA result
Upregulated	Upregulated cont	Upregulated cont	Upregulated cont	Upregulated cont	Downregulated
RPS18	RPL30	RPL3	RPLP2	RPS7	SMARCA4
RPL27	RPL37A	RPL13A	RPL8	RPS3A	EP300
RPS15A	RPS21	SRP9	RPLP0	NUP37	CTBP1
RPS14	RPS11	RPL38	RPL13	NUP54	CARM1
RPS3	RPL35A	RPL18	RPL28	PSMC3	CAV1
RPL24	RPL27A	RPL7A	RPL5	ARL6IP1	BCL2
RPS12	RPL10A	RPL36	RPL32	UBB	NOTCH1
RPS13	RPL12	RPS8	RPS15	PSMB1	CREBBP
RPS5	RPSA	RPL35	RPL23A	TCEB1	SRCAP
RPL26	RPS4X	RPS25	RPL39	PSMA3	HIPK2
RPS6	RPS27A	RPS9	RPS19	RBX1	SLC22A5
RPS20	RPS24	RPL31	RPL11	PSMB3	INHBB
RPS16	RPL9	RPL29	UBA52	EIF4A2	ANK2
RPS18	RPL30	RPL3	RPLP2	PSMA4	SMARCA4
RPL27	RPL37A	RPL13A	RPL8	PSMB4	EP300
RPS15A	RPS21	SRP9	RPLP0	EIF3E	CTBP1
RPS14	RPS11	RPL38	RPL13		CARM1
RPS3	RPL35A	RPL18	RPL28		CAV1
RPL24	RPL27A	RPL7A	RPL5		BCL2
RPS12	RPL10A	RPL36	RPL32		NOTCH1
RPS13	RPL12	RPS8	RPS15		CREBBP
RPS5	RPSA	RPL35	RPL23A		SRCAP
RPL26	RPS4X	RPS25	RPL39		HIPK2
RPS6	RPS27A	RPS9	RPS19		SLC22A5
RPS20	RPS24	RPL31	RPL11		INHBB
RPS16	RPL9	RPL29	UBA52		ANK2

Cytoscape

Though I can run GSEA successfully in docker container, for reasons I don’t know, when I tried to run docker by code in the docker container. It always returns

"Error in curl::curl_fetch_memory(url, handle = handle) :

Failed to connect to localhost port 1234: Connection refused"

So I decided to create the html notebook for cytoscape pipeline outside the container, then merge the resulstant image file directly into the final hrml report. The cytoscape pipeline file are also submitted, named EM pipeline.Rmd. Since the GSEA result created by running code above in Docker container are causing problem becasue of in-Docker file path, I ran the code chunk for GSEA outside of docker first and ran cytoscape use this GSEA result. I manually set the path to GSEA result.

Original EM before Manual Layout

This enrichment map is created with p-value cutoff = 0.005 and FDR q-value cutoff=0.005. This q-value is selected to reduce size of network

This enrichment map includes 224 nodes and 4610 edges. 24 of 356 nodes are isolated nodes.

Red node represent ,and blue nodes represents.

Annotate Network

Used autoannotation for this step. Default parameter were chosen. Detailed parameter information are listed below:

Cluster Source: clusterMaker2
ClusterMaker Algorithm: MCL Cluster
Edge Attribute: EnrichmentMap:: simlilarity_coefficient
Label Maker: WordCloud: Adjacent Words(default)
Max Words Per Label: 3
Word Adjancency Bonus: 8
Normalization Factor: 5
Attribute Names: [EnrichmentMap::GS_DESCR]
Display style: Clustered-Standard
Max Words per Cloud: 250
Cluster Cutoff: 1.0
Min Word Occurrence: 1

Additional Information about the EnrichmentMap:

Size of node correspond to size of geneset. Red nodes correspond to upregulated, blue nodes correspond to downregulated. Labels of node are geneset description. Thickness of the edge correspond to similarity coefficient. The more genes two nodes share in common, the thicker the edge.

Publication-ready figure

Theme Network

Interpretation

Similar to result from assignment 2, GSEA and enrichment map result support the original paper.

Recall the original paper(Petrossian et al. 2018), it says that the ER treatment is important for increased CDK4/6 response in HR+ breast cancer. Refer to our enrichment map, proteasome degradation is an important feature that related to upregulated enrichment result. Proteasome degradation leads to inhibited proteasome activity, which could increase tumour killing by decrease concenration of P-glyco-protein in membrane cells.(Orlowski and Dees 2002) In addition, mitocchondrial translational elongation is also upregulated term. Since mitochodrial dysfunction is associated with increased aggressiveness of breast cancer(Lunetti et al. 2019), upregulated translational elongation is supposed to be able to help breast cancer patient. Moreover, estrogen response, which is positively related to cancer cell growth(Petrossian et al. 2018), is downregulated.

Therefore, the conclusion of the original paper is supported by our analysis.

Specific pathway

Here I chose proteasome degradation. There are 2 reasons for chosing this pathway. First, this is a major pathway that clearly related to increased tumour killing. Second, size of its geneset is relatively large.

Here I use GeneMania, chose automatic network weighting and 0 max resultant gene, as default. Predicted link is the one weight the most in the resultant network, but I chose not to show this type of link since I’m more interested in confirmed relationships. In addition, I removed co-localization links.

legend

I examined the gene node with highest rank, which are UBA7, IFNG, HLA-G, and PSMD5. UBA7 can perform as a marker for breast cancer patient since expression of UBA7 will be significantly reduced in breast cancer(Lin et al. 2020), however, here in our sample, expression of UBA7 is increased.As for IFNG, IFNG helps cancer therapy by facilitate tumor clearance and tumor escape (Ni and Lu 2018), therefore also has a positive effect in breast cancer therapies. HLA-G is rarely found in breast cancer tissues, but here it is highly upregulated(Palmisano et al. 2002), which indicate the experiment treatment is related to positive effect on breast cancer.At, last PSMD5 encodes essential 26s subunit of proteasome, since we observed a highly upregulated score for proteasome degradation, it is reasonable that there is more proteasome synthesised. So result of examination on proteasome degradation is consistent with what I found on enrichment map.

References

Lin, Meng, Yanqing Li, Shanshan Qin, Yan Jiao, and Fang Hua. 2020. “Ubiquitin-Like Modifier-Activating Enzyme 7 as a Marker for the Diagnosis and Prognosis of Breast Cancer.” Oncology Letters 19 (4). Spandidos Publications: 2773–84.

Lunetti, Paola, Mariangela Di Giacomo, Daniele Vergara, Stefania De Domenico, Michele Maffia, Vincenzo Zara, Loredana Capobianco, and Alessandra Ferramosca. 2019. “Metabolic Reprogramming in Breast Cancer Results in Distinct Mitochondrial Bioenergetics Between Luminal and Basal Subtypes.” The FEBS Journal 286 (4). Wiley Online Library: 688–709.

Merico, Daniele, Ruth Isserlin, Oliver Stueker, Andrew Emili, and Gary D Bader. 2010. “Enrichment Map: A Network-Based Method for Gene-Set Enrichment Visualization and Interpretation.” PloS One 5 (11). Public Library of Science.

Ni, Ling, and Jian Lu. 2018. “Interferon Gamma in Cancer Immunotherapy.” Cancer Medicine 7 (9). Wiley Online Library: 4509–16.

Orlowski, Robert Z, and E Claire Dees. 2002. “The Role of the Ubiquitination-Proteasome Pathway in Breast Cancer: Applying Drugs That Affect the Ubiquitin-Proteasome Pathway to the Therapy of Breast Cancer.” Breast Cancer Research 5 (1). Springer: 1.

Palmisano, Giulio Lelio, Maria Pia Pistillo, Paolo Fardin, Paolo Capanni, Guido Nicolò, Sandra Salvi, Bruno Spina, Gennaro Pasciucco, and Giovanni Battista Ferrara. 2002. “Analysis of Hla-G Expression in Breast Cancer Tissues.” Human Immunology 63 (11). Elsevier: 969–76.

Petrossian, Karineh, Noriko Kanaya, Chiao Lo, Pei-Yin Hsu, Duc Nguyen, Lixin Yang, Lu Yang, et al. 2018. “ER\(\alpha\)-Mediated Cell Cycle Progression Is an Important Requisite for Cdk4/6 Inhibitor Response in Hr+ Breast Cancer.” Oncotarget 9 (45). Impact Journals, LLC: 27736.

Subramanian, Aravind, Pablo Tamayo, Vamsi K Mootha, Sayan Mukherjee, Benjamin L Ebert, Michael A Gillette, Amanda Paulovich, et al. 2005. “Gene Set Enrichment Analysis: A Knowledge-Based Approach for Interpreting Genome-Wide Expression Profiles.” Proceedings of the National Academy of Sciences 102 (43). National Acad Sciences: 15545–50.

Assignment 3 - Data set Pathway and Network Analysis

YUhan Hu